Heart Failure Prediction1

age Age

anaemia Decrease of red blood cells or hemoglobin (boolean)

creatinine_phosphokinase Level of the CPK enzyme in the blood (mcg/L)

diabetes If the patient has diabetes (boolean)

ejection_fraction Percentage of blood leaving the heart at each contraction (percentage)

high_blood_pressure If the patient has hypertension (boolean)

platelets Platelets in the blood (kiloplatelets/mL)

serum_creatinine Level of serum creatinine in the blood (mg/dL)

serum_sodium Level of serum sodium in the blood (mEq/L)

sex Woman or man (binary)

smoking If the patient smokes or not (boolean)

time Follow-up period (days)

DEATH_EVENT If the patient deceased during the follow-up period (boolean)

import pandas as pd 
import numpy as np 
import klib 
import seaborn as sns
from colorama import Fore, Back, Style
import pandas_profiling as pp 
import matplotlib.pyplot as plt
from mlxtend.plotting import plot_confusion_matrix
from plotly.offline import plot, iplot, init_notebook_mode
import plotly.graph_objs as go
from plotly.subplots import make_subplots
import plotly.express as px
from statsmodels.formula.api import ols
import plotly.graph_objs as gobj
import plotly.figure_factory as ff
import  warnings
warnings.filterwarnings("ignore")
init_notebook_mode(connected=True)
%matplotlib

Using matplotlib backend: Qt5Agg

# importing dataset 
df = pd.read_csv("heart_failure_clinical_records_dataset.csv")
df

	age	anaemia	creatinine_phosphokinase	diabetes	ejection_fraction	high_blood_pressure	platelets	serum_creatinine	serum_sodium	sex	smoking	time	DEATH_EVENT
0	75.0	0	582	0	20	1	265000.00	1.9	130	1	0	4	1
1	55.0	0	7861	0	38	0	263358.03	1.1	136	1	0	6	1
2	65.0	0	146	0	20	0	162000.00	1.3	129	1	1	7	1
3	50.0	1	111	0	20	0	210000.00	1.9	137	1	0	7	1
4	65.0	1	160	1	20	0	327000.00	2.7	116	0	0	8	1
...	...	...	...	...	...	...	...	...	...	...	...	...	...
294	62.0	0	61	1	38	1	155000.00	1.1	143	1	1	270	0
295	55.0	0	1820	0	38	0	270000.00	1.2	139	0	0	271	0
296	45.0	0	2060	1	60	0	742000.00	0.8	138	0	0	278	0
297	45.0	0	2413	0	38	0	140000.00	1.4	140	1	1	280	0
298	50.0	0	196	0	45	0	395000.00	1.6	136	1	1	285	0

299 rows × 13 columns

profile =pp.ProfileReport(df, title ="Heart Failure Profile", html={"style": {"full_width": True}})
profile

Performing EDA on the datase

df.dtypes

age                         float64
anaemia                       int64
creatinine_phosphokinase      int64
diabetes                      int64
ejection_fraction             int64
high_blood_pressure           int64
platelets                   float64
serum_creatinine            float64
serum_sodium                  int64
sex                           int64
smoking                       int64
time                          int64
DEATH_EVENT                   int64
dtype: object

# plotting and analysis age variables
df_hist = [df["age"].values]
groupBylabels = ['age']

fig = ff.create_distplot(df_hist, groupBylabels)
fig.update_layout(title_text = "Age Distribution plot")

# Creating boxplot for Age variable and setting gender into male= 1, and female = 0
fig = px.box(df, x = "sex", y="age", points="outliers")
fig.update_layout(title = "Male = 1, Female = 0")
fig.show()

male = df[df["sex"]==1]
female = df[df["sex"]==0]

male_survi = male[df["DEATH_EVENT"]==0]
male_not = male[df["DEATH_EVENT"]==1]
female_survi = female[df["DEATH_EVENT"]==0]
female_not = female[df["DEATH_EVENT"]==1]

labels = ['Male - Survived','Male - Not Survived', "Female -  Survived", "Female - Not Survived"]
values = [len(male[df["DEATH_EVENT"]==0]),len(male[df["DEATH_EVENT"]==1]),
         len(female[df["DEATH_EVENT"]==0]),len(female[df["DEATH_EVENT"]==1])]
fig = go.Figure(data=[go.Pie(labels=labels, values=values, hole=.4)])
fig.update_layout(
    title_text="Analysis on Survival - Gender")
fig.show()

Gender factor Analysis

surv = df[df["DEATH_EVENT"]==0]["age"]
not_surv = df[df["DEATH_EVENT"]==1]["age"]
hist_data = [surv, not_surv]
group_labels = ["Survived", "Not Survived"]
fig = ff.create_distplot(hist_data,group_labels, bin_size=0.5)
fig.update_layout(
    title_text = "Analysis in Age on Survival Status")
fig.show()

A close look at this digram aboove shows that the Not survival rate cut cross all groups. And also, survival is high on 40 to 80

fig = px.violin(df, y="age", x="sex", color="DEATH_EVENT", box=True, points="all", hover_data=df.columns)
fig.update_layout(title_text="Analysis in Age and Gender on Survival Status")
fig.show()

Age Report

Survival spread is high in age's flow of 40 to 70 The Survival is high for both male between 50 to 60 and female's age between 60 to 70 respectively

fig = px.violin(df, y="age", x="smoking", color="DEATH_EVENT", box=True, points="all", hover_data=df.columns)
fig.update_layout(title_text="Analysis in Age and Smoking on Survival Status")
fig.show()

• The Survival is high for not smoking person 55 to 65, while for smoking person it is between 50 to 60
• Death event for smoking person is high than not smoking person

fig = px.violin(df, y="age", x="diabetes", color="DEATH_EVENT", box=True, points="all", hover_data=df.columns)
fig.update_layout(title_text="Analysis in Age and Diabetes on Survival Status")
fig.show()

Other factors

Histogram plots

fig = px.histogram(df, x = "creatinine_phosphokinase", color = "DEATH_EVENT", marginal="violin", hover_data = df.columns)
fig.show()

fig = px.histogram(df, x="ejection_fraction", color="DEATH_EVENT", marginal="violin", hover_data=df.columns)
fig.show()

fig = px.histogram(df, x="platelets", color="DEATH_EVENT", marginal="violin", hover_data=df.columns)
fig.show()

fig = px.histogram(df, x="serum_creatinine", color="DEATH_EVENT", marginal="violin", hover_data=df.columns)
fig.show()

fig = px.histogram(df, x="serum_sodium", color="DEATH_EVENT", marginal="violin",hover_data=df.columns)
fig.show()

Distribution plots Survived and Not-Survived

surv = df[df['DEATH_EVENT']==0]['serum_sodium']
not_surv = df[df['DEATH_EVENT']==1]['serum_sodium']
hist_data = [surv,not_surv]
group_labels = ['Survived', 'Not Survived']
fig = ff.create_distplot(hist_data, group_labels, bin_size=0.5)
fig.update_layout(
    title_text="Analysis in Serum Sodium on Survival Status")
fig.show()

surv = df[df['DEATH_EVENT']==0]['serum_creatinine']
not_surv = df[df['DEATH_EVENT']==1]['serum_creatinine']
hist_data = [surv,not_surv]
group_labels = ['Survived', 'Not Survived']
fig = ff.create_distplot(hist_data, group_labels, bin_size=0.5)
fig.update_layout(
    title_text="Analysis in Serum Creatinine on Survival Status")
fig.show()

surv = df[df['DEATH_EVENT']==0]['ejection_fraction']
not_surv = df[df['DEATH_EVENT']==1]['ejection_fraction']
hist_data = [surv,not_surv]
group_labels = ['Survived', 'Not Survived']
fig = ff.create_distplot(hist_data, group_labels, bin_size=0.5)
fig.update_layout(
    title_text="Analysis in Ejaction Fraction on Survival Status")
fig.show()

Pie Charts

labels = ['No Diabetes','Diabetes']
diabetes_yes = df[df['diabetes']==1]
diabetes_no = df[df['diabetes']==0]
values = [len(diabetes_no), len(diabetes_yes)]
fig = go.Figure(data=[go.Pie(labels=labels, values=values, hole=.4)])
fig.update_layout(
    title_text="Analysis on Diabetes")
fig.show()

From the above pie chart i can conclude that in i the dataset 58.2% (174) pepole are No Diabetes Person and 41.8% (125) are Diabetes Person.

fig = px.pie(df, values='diabetes',names='DEATH_EVENT', title='Diabetes Death Event Ratio')
fig.show()

diabetes_yes_survi = diabetes_yes[df["DEATH_EVENT"]==0]
diabetes_yes_not_survi = diabetes_yes[df["DEATH_EVENT"]==1]
diabetes_no_survi = diabetes_no[df["DEATH_EVENT"]==0]
diabetes__no_not_survi = diabetes_no[df["DEATH_EVENT"]==1]

labels = ['Diabetes Yes - Survived','Diabetes Yes - Not Survived', 'Diabetes NO - Survived', 'Diabetes NO - Not Survived']
values = [len(diabetes_yes[df["DEATH_EVENT"]==0]),len(diabetes_yes[df["DEATH_EVENT"]==1]),
         len(diabetes_no[df["DEATH_EVENT"]==0]),len(diabetes_no[df["DEATH_EVENT"]==1])]
fig = go.Figure(data=[go.Pie(labels=labels, values=values, hole=.4)])
fig.update_layout(
    title_text="Analysis on Survival - Diabetes")
fig.show()

• From above pie charts we can conclude that in our dataset diabetes from 174 of Non Diabetes person 118 are survived and 56 are not survived and
• From 125 Diabetes person 85 are survived, while 40 are not survived.

anaemia_yes = df[df['anaemia']==1]
anaemia_no = df[df['anaemia']==0]

labels = ['No Anaemia', 'Anaemia']
values = [len(anaemia_no), len(anaemia_yes)]
fig = go.Figure(data=[go.Pie(labels=labels, values=values, hole=.4)])
fig.update_layout(
    title_text="Analysis on - Anaemia")
fig.show()

From the above pie chart we can conclude that in our dataset 56.9%(170) pepoles are Non anaemic pepoles, while 43.1%(129) are anaemic pepoles.

fig = px.pie(df, values='anaemia',names='DEATH_EVENT', title='Anaemia Death Event Ration')
fig.show()

anaemia_yes_survi = anaemia_yes[df["DEATH_EVENT"]==0]
anaemia_yes_not_survi = anaemia_yes[df["DEATH_EVENT"]==1]
anaemia_no_survi = anaemia_no[df["DEATH_EVENT"]==0]
anaemia_no_not_survi = anaemia_no[df["DEATH_EVENT"]==1]

labels = ['Anaemia Yes - Survived','Anaemia Yes - Not Survived', 'Anaemia No - Survived', 'Anaemia NO - Not Survived']
values = [len(anaemia_yes[df["DEATH_EVENT"]==0]),len(anaemia_yes[df["DEATH_EVENT"]==1]),
         len(anaemia_no[df["DEATH_EVENT"]==0]),len(anaemia_no[df["DEATH_EVENT"]==1])]
fig = go.Figure(data=[go.Pie(labels=labels, values=values, hole=.4)])
fig.update_layout(
    title_text="Analysis on Survival - Anaemia")
fig.show()

• From above pie charts i can conclude that in the dataset diabetes from 170 of Non Anaemic person 120 are survived and 50 are not survived and
• From 129 Anaemic person 83 are survived, while 46 are not survived.

hbp_yes = df[df['high_blood_pressure']==1]
hbp_no = df[df['high_blood_pressure']==0]

labels = ["No High BP","High BP"]
values = [len(hbp_no), len(hbp_yes)]

fig = go.Figure(data=[go.Pie(labels=labels, values=values, hole=.4)])
fig.update_layout(
    title_text="Analysis on - High Blood Pressure")
fig.show()

From the above pie chart we can conclude that in our dataset 64.9%(194) pepoles are Non High BP pepoles, while 35.1%(105) are High BP pepoles.

fig = px.pie(df, values='high_blood_pressure',names='DEATH_EVENT', title='High Blood Pressure Death Event Ratio')
fig.show()

hbp_yes_survi = hbp_yes[df["DEATH_EVENT"]==0]
hbp_yes_not_survi = hbp_yes[df["DEATH_EVENT"]==1]
hbp_no_survi = hbp_no[df["DEATH_EVENT"]==0]
hbp_no_not_survi = hbp_no[df["DEATH_EVENT"]==1]

labels = ['HBP Yes - Survived','HBP Yes - Not Survived', 'HBP No - Survived', 'HBP NO - Not Survived']
values = [len(hbp_yes[df["DEATH_EVENT"]==0]),len(hbp_yes[df["DEATH_EVENT"]==1]),
         len(hbp_no[df["DEATH_EVENT"]==0]),len(hbp_no[df["DEATH_EVENT"]==1])]
fig = go.Figure(data=[go.Pie(labels=labels, values=values, hole=.4)])
fig.update_layout(
    title_text="Analysis on Survival - HBP(high blood pressure)")
fig.show()

• From above pie charts we can conclude that in our dataset diabetes from 194 of Non High BP person 137 are survived and 57 are not survived and
• From 105 High BP person 66 are survived, while 39 are not survived.

smoking_yes = df[df['smoking']==1]
smoking_no = df[df['smoking']==0]

labels = ['No Smoking','Smoking']
values = [len(smoking_no), len(smoking_yes)]

fig = go.Figure(data=[go.Pie(labels=labels, values=values, hole=.4)])
fig.update_layout(
    title_text="Analysis on - Smoking")
fig.show()

From the above pie chart we can conclude that in our dataset 67.9%(203) pepoles are Non Smoking pepoles, while 32.1%(96) are Smoking pepoles.

fig = px.pie(df, values='smoking',names='DEATH_EVENT', title='Smoking Death Event Ratio')
fig.show()

smoking_yes_survi = smoking_yes[df["DEATH_EVENT"]==0]
smoking_yes_not_survi = smoking_yes[df["DEATH_EVENT"]==1]
smoking_no_survi = smoking_no[df["DEATH_EVENT"]==0]
smoking_no_not_survi = smoking_no[df["DEATH_EVENT"]==1]

labels = ['Smoking Yes - Survived','Smoking Yes - Not Survived', 'Smoking No - Survived', 'Smoking NO- Not Survived']
values = [len(smoking_yes[df["DEATH_EVENT"]==0]),len(smoking_yes[df["DEATH_EVENT"]==1]),
         len(smoking_no[df["DEATH_EVENT"]==0]),len(smoking_no[df["DEATH_EVENT"]==1])]
fig = go.Figure(data=[go.Pie(labels=labels, values=values, hole=.4)])
fig.update_layout(
    title_text="Analysis on Survival - Smoking")
fig.show()

From above pie charts we can conclude that in our dataset diabetes from 203 of Non Smoking person 137 are survived and 66 are not survived and From 96 Smoking person 66 are survived, while 30 are not survived.

Heatmap

plt.figure(figsize=(10,10))
sns.heatmap(df.corr(), vmin=-1, cmap='coolwarm', annot=True)

<AxesSubplot:>

Data Modeling

Train Test Split

Features = ['time','ejection_fraction','serum_creatinine']
X = df[Features]
y = df["DEATH_EVENT"]

from sklearn.model_selection import train_test_split
X_train, X_test, y_train, y_test = train_test_split(X, y, train_size=0.7, random_state=123)

from sklearn.linear_model import LogisticRegression
from sklearn.svm import SVC
from sklearn.tree import DecisionTreeClassifier
from sklearn.neural_network import MLPClassifier
from sklearn.ensemble import RandomForestClassifier
from xgboost import XGBClassifier
from catboost import CatBoostClassifier
import lightgbm
import xgboost
from sklearn.ensemble import GradientBoostingClassifier
from sklearn.naive_bayes import GaussianNB
from sklearn.neighbors import KNeighborsClassifier
from sklearn.metrics import accuracy_score
from sklearn.metrics import confusion_matrix

names = [
    "CatBoostClassifier",
    "Logistic Regression",
    "Support Vector Machine",
    "Decision Tree",
    "Neural Network",
    "Random Forest",
    "XGBoost",
    "LGBMClassifier",
    "XGBRFClassifier",
    "GradientBoosting",
    "GaussianNB",
    "KNeighborsClassifier"
]
models = [
    CatBoostClassifier(verbose= False),
    LogisticRegression(),
    SVC(),
    DecisionTreeClassifier(),
    MLPClassifier(),
    RandomForestClassifier(),
    XGBClassifier(),
    lightgbm.LGBMClassifier(max_depth=2, random_state=4),
    xgboost.XGBRFClassifier(max_depth=3, random_state=1),
    GradientBoostingClassifier(max_depth=2, random_state=1),
    GaussianNB(),
    KNeighborsClassifier(n_neighbors=5, p=2, metric='minkowski')
]

accuracy=[]
for model, name in zip(models,names):
    model.fit(X_train, y_train)
    
    y_pred = model.predict(X_test)
    print('Confusion matrix of ',name)
    print(confusion_matrix(y_test, y_pred))
    ac = accuracy_score(y_test, y_pred)
    print('Accuracy score is ',ac)
    accuracy.append(ac)
    print('='*50)

Accuracy_list = pd.DataFrame(list(zip(names, accuracy)),columns =['Model', 'Accuracy'])
Accuracy_list= Accuracy_list.sort_values('Accuracy', axis=0, ascending=False, inplace=False, kind='quicksort', na_position='last', ignore_index=True, key=None)

Confusion matrix of  CatBoostClassifier
[[49  4]
 [15 22]]
Accuracy score is  0.7888888888888889
==================================================
Confusion matrix of  Logistic Regression
[[49  4]
 [17 20]]
Accuracy score is  0.7666666666666667
==================================================
Confusion matrix of  Support Vector Machine
[[52  1]
 [20 17]]
Accuracy score is  0.7666666666666667
==================================================
Confusion matrix of  Decision Tree
[[46  7]
 [17 20]]
Accuracy score is  0.7333333333333333
==================================================
Confusion matrix of  Neural Network
[[53  0]
 [37  0]]
Accuracy score is  0.5888888888888889
==================================================
Confusion matrix of  Random Forest
[[49  4]
 [14 23]]
Accuracy score is  0.8
==================================================
Confusion matrix of  XGBoost
[[48  5]
 [16 21]]
Accuracy score is  0.7666666666666667
==================================================
Confusion matrix of  LGBMClassifier
[[50  3]
 [13 24]]
Accuracy score is  0.8222222222222222
==================================================
Confusion matrix of  XGBRFClassifier
[[51  2]
 [16 21]]
Accuracy score is  0.8
==================================================
Confusion matrix of  GradientBoosting
[[50  3]
 [14 23]]
Accuracy score is  0.8111111111111111
==================================================
Confusion matrix of  GaussianNB
[[52  1]
 [21 16]]
Accuracy score is  0.7555555555555555
==================================================
Confusion matrix of  KNeighborsClassifier
[[51  2]
 [17 20]]
Accuracy score is  0.7888888888888889
==================================================

plt.rcParams['figure.figsize']=20,6 
sns.set_style("darkgrid")
ax = sns.barplot(x = 'Model',y = 'Accuracy',data = Accuracy_list , palette = "rocket", saturation =1.5)
plt.xlabel("Model", fontsize = 20 )
plt.ylabel("Accuracy", fontsize = 20)
plt.title("Accuracy of different Models", fontsize = 20)
plt.xticks(fontsize = 11, horizontalalignment = 'center', rotation = 8)
plt.yticks(fontsize = 13)
for p in ax.patches:
    width, height = p.get_width(), p.get_height()
    x, y = p.get_xy() 
    ax.annotate(f'{height:.2%}', (x + width/2, y + height*1.02), ha='center', fontsize = 'x-large')
plt.show()